Topic prominence is not a factor of variation between. Brazilian and European Portuguese. João Costa. FCSH - Universidade Nova de Lisboa.

Transcription

1 Topic prominence is not a factor of variation between Brazilian and European Portuguese João Costa FCSH - Universidade Nova de Lisboa Abstract The existing syntactic differences between European and Brazilian Portuguese have received much attention in recent literature on generative syntax. According to several authors, Brazilian Portuguese became a discourse-oriented language with topic-prominence. This article shows that many constructions used to illustrate the specificity of Brazilian Portuguese as topic-prominent can also be found in European Portuguese. Accordingly, it is argued that the main difference between the two varieties of the language is not in discourse, but in the abstract syntactic properties associated with Inflection. This argument contributes to dispensing with discourse primitives in the syntactic component. Keywords: variation, topic, agreement, parameter, subject 1. Introduction. Understanding the variation between two closely related grammars like European Portuguese (EP) and Brazilian Portuguese (BP) opens up a great

2 window into the comprehension of factors underlying parametric microvariation. As emphasized by Galves (2001), the fact that EP and BP are so close makes the comparative study of these two grammars particularly revealing, both diachronically and synchronically. Recent studies on BP reveal many syntactic aspects in which BP behaves differently from EP, in particular in what concerns the syntax of subjects (Duarte 1993 and subsequent work), and in what appears to be a differentiated behaviour of topics (Pontes 1987). The observation of some of these differences lead some authors to propose that BP has turned into a topic-prominent language in the sense of Li and Thompson (1976), acquiring the status of discourse-oriented language (Duarte and Kato 2008). According to this view, the ongoing change in BP would instantiate a case of macro-parametric change, in the sense of Baker (1995), since it would be a case in which not only a small detailed feature of the language would have changed, but a set of changes allowing for a typological distinction, implying major differences in a whole array of sub-parameters and properties. Even if one is sceptical about the existence of macroparameters, this change would be an interesting study case for a typological difference between languages that do or do not encode discourse primitives in the syntactic component. The goal of this paper is to question whether the difference between BP and EP can be explained in terms of a change in what concerns discourse orientation. The argument will be built around two major axes:

3 on the one hand, I will show that some of the constructions used to illustrate the changes in BP are also available in EP; on the other hand, I will argue that some of the data do not constitute sound evidence for a characterization in terms of topic-prominence. I will further show that some of the major differences between the two varieties of the language are based on behaviour of agreement, which will indicate that the syntactic change has to do with properties of Infl and not with macro-parametric aspects. Finally, I will address some issues regarding the need for integrating syntactic primitives in the syntactic component. The article is organized as follows: in section 2, the main arguments put forward in the literature in favor of the idea that BP is a discourseoriented language are presented; in section 3, data from EP are presented, casting doubt on a difference between the two grammars in what concerns discourse orientation. It is also shown that the main differences have to do with the agreement system. Section 4 discusses some arguments for a more autonomous view of syntax, discussing the locus of syntactic varation. Section 5 presents the article s main conclusions. 2. BP as a discourse oriented language. It is undeniable that EP and BP currently diverge in a substantial number of syntactic aspects. Some of the major contributions for the comparative study of the two grammars in the framework of Principles and

4 Parameters are due to Galves (2001), summarizing many of the relevant differences, and to Duarte (1993), who revealed the tendency of BP towards a significant reduction in the percentage of referential null subjects, and a corresponding increase in the production of overt referential pronouns. However, it is not possible to say that BP became a non null subject language of the English or French type, since it still allows for null subjects in several contexts. This is the case for expletive null subjects or subjects controlled by a discourse topic (Figueiredo Silva 1996). Pontes (1987) observes that in BP, unlike in EP, one can find sentences like (1), in which a topic triggers verbal agreement: (1) Essas casas batem sol. Those houses hit-3pl sun Sun hits those houses. The observation that verbal agreement may be triggered by a non-subject topic led Pontes to suggest that BP turned into a topic-prominent language in the sense of Li and Thompson (1976). According to these authors, a topic-prominent language differs from a subject-prominent language in that most syntactic processes are sensitive to discourse functions (like topic) and not to argument functions (like subject). This hypothesis was largely explored in the work of E. Duarte, who extends it to other constructions and adopts it as the solution to explain the

5 fact that BP is not becoming a non null subject language of the English or French type, in spite of the loss of referential null subjects. Duarte (2004) and Duarte and Kato (2008) list the following constructions as arguments in favour of characterizing BP as a topic language: a) Construction 1: Free occurrence of double subjects. As (2) illustrated, BP allows subject doubling by a pronoun. This is reminiscent of French subject left-dislocation, described in Rizzi (1986), and illustrated in (3): (2) a. [Essa competência] i, ela i é de natureza mental. That competence she is of nature mental That competence is mental in nature. b. [Mulher nenhuma] i ela i pode querer dominar o homem. [O homem] i ele i é livre por natureza. [A mulher] i ela i tem que aceitar isso. woman none she may want dominate the man the man he is free by nature the woman she has that accept that No woman may want to dominate the man. Men are free by nature. Women must accept that. c. [Toda criança] i ela i aprende rápido a gostar de coca-cola.

6 Every child she learns quicly to enjoy coca-cola Every child quickly learns to enjoy coke. d. [O que é bom, o que é de qualidade] i ele i fica; [o que é ruim] i ele i se what is good, what is of quality he stays; what is bad he SE perde. looses What is good and has quality remains; what is bad gets lost. (3) Jean, il est là. Jean, he is there Jean is there. Note that this doubling construction has peculiar properties, since it is allowed with inanimate DPs, as in (2a), with generic DPs, as in (2b), with quantified DPs, as in (2c), and with free relatives (1d). I will return to this observation below. b) Construction 2: overt locative and deictic subjects (Duarte 2004). In spite of the fact that the gradual loss of referential null subjects in BP is not mirrored by a diminution of expletive null subjects, Duarte (2004)

7 observes that there are many cases in which the subject position of weather verbs is filled in by a lexical locative or deictic subject, as shown in (4). It is clear that it is the subject position that is occupied, because there are cases in which overt agreement obtains between the lexical locative subject and the verb, as in (4c): (4) a. Aqui tá quente. Here is hot It s hot in here. b. São Paulo chove; o Rio faz sol. São Paulo rains; the Rio makes sun It rains in São Paulo; it s sunny in Rio. c. Essas florestas chovem muito. Those forests rain-3pl a lot It rains a lot in those forests. d. Petrópolis, aquilo chove demais. Petrópolois, that rains too much It rains too much in Petrópolis. According to Duarte (2004), all these cases instantiate configurations in which a topic is attracted to the highest specifier position in the clause, which can be, in some cases, the subject position.

8 c) Construction 3: Existential construction with pronoun insertion. Several authors have shown that the verb ter/have with existential value, which, in a pro-drop language, would have an expletive null subject, may occur with a preverbal pronoun filling in the subject position (Kato and Tarallo 1986, Duarte 2004, Callou and Duarte 2005, Callou and Avelar 2007). The sentences in (5) illustrate this possibility: (5) a. Cê tem prédios lindos em Londres. You have buildings beautiful in London There are beautiful buildings in London. b. Eu tenho uma papelaria ali na esquina que tira cópia baratinho. I have-1sg a stationery there on the corner that takes xerox cheap There s a stationery around the corner with cheap xerox. c. A gente não tem mais comércio no centro da cidade. The people not has more commerce in the centre of the town There s no longer any commerce in the town centre. According to most analyses, the inserted pronoun fills in the subject position, but it is simultaneously interpreted as a discourse topic.

9 d) Construction 4: Subject hyper-raising with the verb parecer /seem (Ferreira 2000, among others). As shown in (6), BP permits constructions in which the embedded subject of a finite complement clause is raised to the matrix subject position of the verb parecer/seem, which is not expected according to the case-based classical analysis of raising constructions (Ferreira 2000): (6) a. Tem ocasiões que eu i nem pareço [que t i sou brasileiro]. has occasions that I not-even seem-1sg that am-1sg Brazilian There s occasion in which I don t seem to be Brazilian. b. [Caso [as aulas] i pareçam [que t i vão voltar ]]eu mando eles pra escola de novo case the classes seem-3pl that go-3pl return I send them to school again In case the classes seem to restart, I send them to school again. c. Quando eu brigo, eu pareço [que eu vou explodir de raiva]. When I fight I seem-1sg that I go-1sg burst of rage When I fight, I seem to burst in rage. d. Vocês parecem [que vocês não pensam na vida]. You seem-3pl that you not think-3pl in the life

10 You seem not to think of life. Duarte and Kato (2008) explicitly argue that this construction derives from the discourse configurational nature of BP s grammar, particularly reflecting the fact that the language is topic-prominent. e) Construction 5: Transitive-ergative pairs. Another construction generally listed as part of the evidence to treat BP as a topic-prominent language is the ergatization of transitive verbs (Duarte and Kato 2008). In these cases, the complement of a transitive verb is promoted, and becomes the subject of an ergative verb, as shown in (7): (7) a. A revista tá xerocando. The journal is xeroxing Someone is making copies of the journal. b. Com a reforma, meu jardim destruiu inteirinho. With the reform, my garden destroyed entirely With the reform, my garden got completely destroyed. f) Construction 6: genitive raising in unaccusative constructions.

11 Finally, BP allows for sentences like those illustrated in (8), in which a genitive raises to subject position of an unaccusative verb, because it is a topic. When the topic genitive raises, it fills in the subject position controlling verbal agreement: (8) a. [Meu carro] i furou [o pneu t i ] My car emptied the tyre My car has got a flat tyre. b. [Minhas pernas] i racharam [a pele t i ]. my legs broke-3pl the skin The skin of my legs broke. It is the set of these six properties, rather than each of them isolated, that led the authors referred so far to propose that BP, unlike EP, underwent a change from subject-prominent language to a topic-oriented language. In this sense, there would be a macro-parametric variation between the two varieties of Portuguese, in the sense of Baker (1995). This would place the two languages in different typological groups. A note of caution is due here. Duarte and Kato (2008) raise the hypothesis that, in fact, BP has a mixed status as both subject-prominent and topic-prominent. This analysis tries to ensure that, in spite of the parametric change, it is possible to explain the properties of syntactic configurationality one can detect in the grammar of BP. However, I think

12 this hypothesis deserves further testing. There will be cases in which neutralization emerges, provided syntactic and discourse functions compete for the same position of for affecting a given syntactic process. In such cases, it is expected that the default option of the language becomes evident, in a procedure akin to emergence of the unmarked, as predicted in Prince and Smolensky s (1993) Optimality Theory. Such cases of competition were partly discussed in Costa and Figueiredo Silva (2006). Independently of these issues, if one takes into consideration the six properties listed above, taken from the literature and generally given as allowing for a characterization of BP as a topic-prominent language, one expects that these six properties differerentiate BP from EP. 3. BP s topic-prominent properties in EP. Before starting to evaluate the empirical robustness of the claim that some constructions are specific to BP, I would like to comment the validity of the claim that subject doubling in BP is a good diagnostic for characterizing it as a topic-prominent language. The rationale underlying this idea is that BP would be a language massively resorting to a leftdislocation strategy. Accordingly, this representation for subjects would be a consequence of a resetting of the null subject parameter, which would cause the visibility of topicalized subjects. This would be in line with the view advocated in Barbosa (1995), among others, that preverbal subjects

13 are left-dislocated in null subject languages. BP would, then, illustrate a case in which the clause internal position of the subject is a pronoun, but, given the loss of null subjects, the pronoun would have lexical content. However, as defended in Silva (2004) and Costa, Duarte and Silva (2006), there are critical differences between left dislocation and subject doubling in BP. As a baseline for comparison, let us take the properties of French subject doubling via left dislocation, as described in De Cat (2003). Following this description, Costa, Duarte and Silva (2006) conclude that subject doubling in BP cannot be analyzed as an instance of topic promotion, unlike what happens in French. Citing some of their arguments, whereas in French, one can observe that the antecedent of the pronoun cannot be new information, in BP a doubled subject may occur in out-ofthe-blue contexts, which shows that it does not have the status of a topic. Likewise, whereas in French a quantified subject cannot be doubled by a pronoun (cf. 8), in BP such a constraint is unattested, as the data in (2) revealed: 1 (8) *Chaque enfant, il apporte son livre à l'école. Each child, he takes his book to the school Each child takes his own book to school. 1 There is debate regarding the nature of subject doubling in French. For arguments for the left-dislocation analysis (and against the agreement analysis), see de Cat (2005).

14 In line with the observation that the subjects that serve as antecedents for doubling pronouns in BP do not have the referential properties of elements occurring in left dislocation constructions, which does not confirm the idea that there is topic-prominence, Costa, Duarte and Silva (2006) propose that the pronoun occurring in double subject constructions in BP is a spell-out of the person feature, which has been progressively disappearing from verbal inflection. A corollary of this analysis is that subject doubling cannot be taken as a marker of topic-prominence, since the pronoun would be a feature encoding person, and not topicality. Accordingly, the difference between the two grammars is not in the topic-promotion mechanism, but rather in the way the person feature manifests itself: in Infl in EP and spelled out as a free pronoun in BP. Let us now consider the remaining constructions. If, as was mentioned above, BP differs from EP in that the former the former turned into a discourse oriented language with topic prominence, it is expected that the constructions that characterize BP as such are not attested in the grammar of EP. We will see that this prediction is not borne out, and that some of the constructions described are not exclusive of BP. Since construction 1 (subject doubling by a pronoun) was discarded, given its irrelevance for attesting topic prominence, it will not be considered. a) Construction 2: overt locative and deictic subjects (Duarte 2004).

15 Just like BP, EP allows for filling in the preverbal position of weather verbs with overt locatives or deictic expressions. The examples in (10) contain two cases taken from the internet: (10) a. O Largo Camões, aquilo faz um frio The square Camões, that makes a cold It s very cold at Camões Square (21/07/09) b. Podem-me chamar de parva, de burra, de tudo, por ter voltado a Idanha May me call of dumb, of donkey, of everything, for have returned to Idanha (sim porque aquilo faz um calor desgraçado)! yes because that makes a heat disgraced You may call me dumb, stupid or whatever, for having gone back to Idanha. Yeah, it s terribly hot there. (21/07/09) Unlike what one might suppose, these examples are not infrequent. It is easy to construct examples clearly judged grammatical by any native speaker of EP, like those in (11):

16 (11) a. Aqui está calor. Here is hot It s hot in here. b. O Barlavento faz mais vento. The Barlavento makes more wind It s more windy in Barlavento. In light of these data, it is legitimate to conclude that EP also has overt locative and deictic subjects. That is, construction 2 also exists in this variety of Portuguese. In contrast, what appears to be truly out in EP is the type of construction illustrated for BP in which the locative subject triggers plural agreement on the weather verb, as in the examples *Estas florestas chovem muito/ these forests rain-3pl a lot. As such, it is legitimate to ask what makes the two varieties differ: the filling in of the preverbal position or the agreement system? At first sight, it appears to be the case that the agreement system is the crucial difference, since the two grammars do not differ in what regards the strategy for filling in the preverbal position with overt locative and deictic expressions. b) Construction 3: Existential construction with pronoun insertion. I. Duarte et al. (2002), in a paper on the internal feature specification of pronouns, show that most personal pronouns in EP may be

17 used with arbitrary values. Such arbitrary uses of pronouns are not only possible in generic sentences, as in (12), but also in existential constructions akin to those described for BP, as illustrated in (13): (12) A gente tenta sempre fugir ao fisco. The people tries always to escape to-the taxes We always try to escape taxes. (13) a. A gente tem uma boa padaria no bairro. The people has a good bakery in-the neighborhood There s a good bakery in our neighbourhood. b. Nós temos muita corrupção no país. We have much corruption in the country There s a lot of corruption in our country. c. Tu tens muitos perigos em Setúbal. You have many dangers in Setúbal There are many dangers in Setúbal. d. Eu tenho um aeroporto perto de casa e não consigo dormir. I have an airport close to home and not can sleep There s an airport close to my house, and I can t sleep. In all the cases listed in (13), the sentences could be paraphrased by an existential construction with the verb haver and a null subject. Accordingly,

18 it is possible to conclude that, just like BP, EP has existential constructions with overt pronouns. As such, this domain does not illustrate any remarkable difference between the two varieties of Portuguese. c) Construction 4: Subject hyper-raising with the verb parecer /seem Costa and Rooryck (1995) propose an analysis for EP data, which they treat as a case of pseudo-raising. Such data are not different from what has been called hyper-raising in the literature on BP syntax: (14) a. Eu pareço que estou cansado. I seem-1sg that am-1sg tired I seem to be tired. b. Tu pareces que estás parvo. You seem-2sg that are-2sg fool You seem to be a fool. c. Nós parecemos que estamos doentes. We seem-1pl that are-1pl tired We seem to be tired. This construction also involves a matrix subject associated to parecer /seem that triggers agreement on this verb, in a context in which it appears to have been moved out of the embedded finite complement clause.

19 This type of construction is highly productive in EP. In (15), some examples taken from the internet are given illustrating the productivity of pseudo/hyper-raising: (15) a. Não sei de que região és natural, mas pareces que estás numa pega. Not know of what region are-2sg natural, but seem-2sg that are-2sg in a bullfight I don t know what region you come from, but you seem to be at a bullfight. b. As pessoas dizem que eu nem pareço que tenho 17 anos. The persons say that I not seem-1sg that have-1sg 17 years People say I don t seem to be 17 years old. c. Escrever é bom, sobretudo quando não temos alguém em quem confiar, write-inf is good, mostly when not have someone in whom trust-inf quando estamos sozinhos e parecemos que vamos rebentar e então when are alone and seem-1pl that go-1pl burst and then explodimos

20 explode-1stpl Writing is good, mostly when we have no one we can trust, when we are alone, and we seem to be bursting, and then we explode. d. oh esquece eu às vezes até pareço que como palha com um garfo. Oh forget I at times even seem-1sg that eat-1sg straw with a fork Oh, forget it. Sometimes, I seem to be really stupid. According to Costa and Rooryck (1995), this construction in EP is subject to a couple of aspectual restrictions. In future work, it would be interesting to explore if the same constraints apply in BP. For now, for the purposes of this paper, it is enough to emphasize the fact that pseudo/hyper-raising does not appear in BP only, but also in EP. e) Construction 5: Transitive-ergative pairs. The ergatization of transitive verbs is not exclusive of BP either. The data under (16) illustrate several cases of ergative uses for transitive verbs:

21 (16) a. O trabalho está a imprimir. The work is at print-inf The paper is printing. b. O bolo está a cozer. The cake is at bake-inf The cake is baking. c. Com o calor, a manteiga derreteu todinha. With the heat, the butter melted all With the heat, the butter melted completely. d. A fábrica fechou com a crise. The factory closed with the crisis The factory closed down because of the crisis. e. A cadeira baloiçou com o vento. The chair swung with the wind The wind made the chair swing. These transitive/ergative pairs are well attested in EP (Fiéis 2003, among others), and do not seem to be an innovation in the language, since, according to Fiéis (2003), they are documented since the earliest stages of Portuguese. f) Construction 6: genitive raising in unaccusative constructions.

22 In this last construction, we can finally find a real difference between the two varieties of Portuguese. In EP, sentences like those presented under (17) are ungrammatical: 2 (17) a.??*[o meu carro] i furou [o pneu t i ] My car emptied the tyre My car has got a flat tyre. b. *[As minhas pernas] i racharam [a pele t i ]. my legs broke-3pl the skin The skin of my legs broke. b. *Essas casas batem sol. Those houses hit-3pl much sun The sun hits those houses. c. *Os meus vizinhos morreram a mãe. My neighbors died-3pl the mother My neighbors mother died. d. *As minhas duas árvores apodreceram a raiz. My two trees rotten-3pl the root The root of my two trees got rotten. 2 Interestingly, no Brazilian Portuguese speaker consulted accepts sentences (17c) or (17d). Note that these two sentences are perfectly parallel to the one in (8b). This casts doubt on the productivity of this structure even in BP. For EP, only one instance of this structure was found on the internet (cf. i). Given its oddity and low frequency, this occurrence was not considered to be relevant in the language. (i) O meu carro avariou o elevador do vidro eléctrico My car broke the elevator of-the window electrical The elevator of the electrical window of my car broke. I will have to leave the issue of the productivity of this structure in BP for future research.

23 In contrast with the ungrammaticality of the sentences in (17), it is important to observe that preposing structures in which a topic PP is fronted with an empty P are possible in EP (I. Duarte 1987, 1996). Such topicalizations, named in the literature on Portuguese topicalização selvagem (lit: wild topicalization) are illustrated in (18): (18) a. Essas casas, bate imenso sol. Those houses hit-3sg much sun The sun hits those houses. b. Os meus vizinhos, morreu a mãe. My neighbors died-3sg the mother My neighbors mother died. c. As minhas duas árvores, apodreceu a raiz. My two trees rotten-3sg the root The root of my two trees got rotten. d. O meu carro, furaram os pneus. The my car, emptied-3pl the tyres My car s tyres got flat. e. As minhas pernas, rachou a pele. The my legs, cracked the skin My legs skin cracked.

24 A comparison between the data in (18) and the possible structures in BP allows for an identification of the minimal factors of variation. Crucially, the two grammars EP and BP are not distinguishable in what concerns the strategy of topic promotion, since both make use of syntactic mechanisms for preposing the topic into the left periphery of the clause. The crucial difference between the two grammars is the agreement system, like was mentioned above. Unlike what happens in EP, in BP the verb may agree with the topic. Accordingly, one may raise the (descriptive) hypothesis that the real difference between the two grammars is in the agreement system, as proposed in Costa and Galves (2002). In Costa and Galves (2002), it is proposed that, in Brazilian Portuguese, topics can enter into agreement relations with the verb because the higher projection of the Infl domain is empty, differently from what happens in European Portuguese, in which the same projection is occupied by the subject. So far, it is possible to draw a partial conclusion from the data presented. Out of the six structures that would distinguish BP from EP, characterizing the former as a topic-prominent language, it is was possible to conclude that one of them was not a good trace of topic-prominence, since it involves the doubling of DPs that do not have the referential properties of topics; in addition, it was possible to observe that four out of the six structures do not distinguish the two grammars, since EP has all the constructions that were expected to emerge only in BP. Therefore, out of

25 the six structures, only one can actually distinguish the two varieties: the raising of genitives in unaccusative environments. However, in the analysis of this structure, it was possible to see that the agreement process is more relevant than the topic promotion strategy, since the latter does not differentiate the two varieties of Portuguese. The variation is therefore residual. Nevertheless, it is important to know how to account for it. 4. The locus of variation. The main conclusion of the previous section is that EP does not differ from BP in most of the structures typically listed as characterizing BP as a discourse-configurational language with topic prominence. Given this observation, there are two possible ways to understand the similarities and differences between the two grammars of Portuguese. The first option would be to say that EP and BP do not differ in the structures described in the previous section, because EP is a language with topic prominence as well. However, this does not appear to be an interesting option for several reasons. First, the whole discussion on variation would become empty, since this would mean that the two grammars are identical, now putting the focus on the similarities and annulling the differences. Second, it is possible to show that discourse configurationality is a very limited tool to describe language variation

26 (Costa 2010). For instance, in a comparison between EP and BP in what concerns the distribution of arguments in informational focus contexts, Costa e Figueiredo Silva (2006) show that there is no advantage in including discourse notions in the set of syntactic primitives, since that brings no empirical gains on descriptive grounds. Costa e Figueiredo Silva (2006) argue that the variation in the distribution of pre- and postverbal subjects in EP and BP is primarily conditioned by the syntax of null subjects and not by discourse constraints codified in the syntactic component. They further contend that scrambling of internal arguments does not distinguish the two grammars, neither do ellipsis or null object constructions. All these constructions involve sensitivity to topichood. Moreover, they show that, in both grammars, metrical invisibility (or stress shift) is used as a last resort strategy. These observations converge on two results: on the one hand, there are no significant differences between the two varieties of Portuguese in what concerns discourse-orientation; on the other hand, the availability of last resort strategies indicates that the syntactic component does not include discourse primitives. In other words, syntax manipulates syntactic tools and generates word order that can be interpreted by discourse postsyntactically. The model I advocate here is, therefore, one in which syntax works in a pretty autonomous way. Like Vallduví (1992) proposes, the discourse component acts as an autonomous module, fed by the structures generated

27 by the syntactic component. This discourse-free syntax ensures, in accordance to Newmeyer (2007), that one is able to adhere to some of the basic principles of generative syntax. According to such principles, syntax is autonomous and syntactic operations are not constrained by discourse information. In short, the point I am trying to make is that it is not expected that the locus of variation between EP and BP is located in discourse, since it does not favour the autonomy of syntax and, more importantly, it is empirically inadequate. At this point, it is important to propose an alternative explanation to account for the variation observed in the previous section. Recall that it was possible to see that the main differences between BP and EP had not to do with the strategies for promoting topics. In both grammars, topics may be promoted through preposing strategies. What really looks different is the agreement system, which, in the BP grammar, allows the establishment of agreement between a topic and the verb. Such a possibility is unattested in EP. Further recall that the agreement system, in particular the weakening of the inflection distinctions, was held responsible for the emergence of pronouns in the subject doubling constructions of BP, as a compensatory strategy to spell out person features. As such, the real differences listed in the previous section pin down to one category and its properties: Inflection. This conclusion is a common

28 desire in current generative syntax: locating syntactic variation in precise functional heads or, at least, in some of their properties (Chomsky 1995). For the case at stake in this paper, the crucial factor underlying variation is micro-parametric and strictly syntactic: the properties of Infl, which, having undergone change, determine the type of inflectional morphology and the mechanisms for licensing pronominal subjects. Extending this line of reasoning, Infl determines the mechanisms associated to the licensing of lexical subjects, and the set of categories that it agrees with. Locating the actual source of variation in a strictly syntactic microparameter bears some important consequences. It is now expected that the onset of the changes that determine the differences between EP and BP grammar may be found internally to the syntactic component. Ribeiro (2008) shows, on the basis of 19 th century popular Brazilian Portuguese data, that the loss of referential null subjects can be dissociated from the weakening of verbal morphology. This observation is based on data attesting a low rate of null subjects in a speaker that has not yet lost verbal morphology. Instead, one may conjecture that the weakening of verbal morphology is a consequence of the loss of null subjects. Likewise, out of the several properties linked to the resetting of the null subject parameter, it is known that the loss of VS orders for encoding information focus is very robust (Berlinck 1995). More so than the weakening of inflection morphology, which is subject to variation across dialects and BP speakers.

29 This also signals that the syntactic change is steadier than the morphological change. All these data lead to one conclusion: syntactic variation between EP and BP will be better understood if one looks for the syntactic microparameters explaining the differences between the two grammars. For the case at stake, such a micro-parameter can be related to the properties of Infl. 5. Conclusions. At the beginning of this paper, it was observed that a fine description of the similarities and differences between EP and BP may reveal areas of syntactic micro-variation. Such micro-variation may, however, be formally though of as involving macro-parameters placing languages in different typological families or as involving syntactic micro-parameters reducing crosslinguistic variation to detailed aspects. The evidence put forward in this paper favors the view of abstract micro-parameters, without appeal to grammar components other than syntax. In particular, I argued for the advantages of explaining the differences between EP and BP without integrating discourse primitives in the syntactic component. The data presented in the paper favor the more traditional generative view of syntax as an autonomous component. As a consequence, syntactic

30 change and variation do not necessarily involve variables external to syntax. Syntactic differences may be due to or triggered by - strictly syntactic aspects. References. Baker, Mark (1995). The polysynthesis parameter. New York: Oxford University Press Berlinck, Rosane (1995). La Position du sujet en Portugais. Leuven, Katholieke Univ. E UNICAMP, Ph.D. Dissertation. Callou, Dinah. & Juanito Avelar. (2007). Sobre a emergência do verbo possessivo em contextos existenciais na história do português. In: A. Castilho; M. A. Torres-Morais; M. A.; LR. Lopes; S. Cyrino. (Orgs.) Descrição, história e aquisição do português brasileiro Campinas Callou, Dinah & Maria Eugênia Duarte (2005). A fixação do verbo ter em contextos existenciais. Textos Seleccionados. 20º Encontro da APL, APL, Lisboa, APL: pp de Cat, Cécile (2005). French subject clitics are not agreement markers. Lingua 115: Chomsky, Noam (1995). The Minimalist Program. Cambridge: MIT Press Costa, João (2010). Prosodic proeminence: a syntactic matter? In N.

31 Erteschik-Shir & L. Rochman (eds.) The sound pattern of syntax. New York: Oxford University Press Costa, João & Maria Cristina Figueiredo Silva (2006) On the (in)dependence relations between syntax and pragmatics. In: Valéria Molnár and Susanne Winkler (eds.), The Architecture of Focus, Berlin: Walter de Gruyter. Costa, João & Charlotte Galves (2002). External subjects in two varieties of Portuguese. In: Beyssade, C; Bok-Bennema, R; Drijkoningen, F; Monachesi,P.. (Orgs.). Romance Languages and Linguistic Theory Amsterdam: John Benjamins, p Costa, João & Johan Rooryck (1995). On Pseudo-Raising in English and Portuguese. In L. Nash et al (orgs.) Proceedings of Langue et Grammaire 2, Paris VII, 1997, pp Costa, João, Inês Duarte & Cláudia R. Silva (2006) Construções de redobro em português brasileiro: sujeitos tópicos vs soletração do traço pessoa. Revista Leitura, n. 33. jul./dez.maceió: Imprensa Universitária De Cat, Cécile (2003). French dislocation. Doctoral dissertation, University of York Duarte, Eugênia L.(1993). Do pronome nulo ao pronome pleno: a trajetória do sujeito no português do Brasil. In: I. Roberts & M. A. Kato (orgs.) Português Brasileiro: uma viagem diacrônica Campinas: Ed. da UNICAMP.

32 Duarte, Eugênia L. (2004) On the embedding of a syntactic change. LanguageVariation in Europe: Papers from ICLaVE Uppsala, Sweden: Universitetstryckeriet. Duarte, Eugênia L. (2007). Sobre outros frutos de um projeto herético: o sujeito expletivo e as construções de alçamento. In: A. Castilho; M. A. Torres-Morais; M. A.; LR. Lopes; S. Cyrino. (Orgs.) Descrição, história e aquisição do português brasileiro Campinas: Pontes. Duarte, Eugênia L. & Mary Kato (2008). Mudança paramétrica e orientação para o discurso. Comunicação apresentada no XXIV Encontro Nacional da Associação Portuguesa de Linguística, Braga Duarte, Inês (1987). AConstrução de Topicalização na Gramática do Português: Regência, Ligação e Condições sobre Movimento. Lisboa, Tese de doutoramento. Faculdade de Letras da Universidade de Lisboa Duarte, Inês (1996) A topicalização em português europeu: uma analise comparativa. In: Duarte, I. & Leiria, I. (orgs.). Congresso Internacional sobre o Português, Actas , Lisboa: Colibri Duarte, Inês, M. João Freitas, Anabela Gonçalves, Matilde Miguel & Celeste Rodrigues (2002). "Geometria de traços e distribuição de pronomes sujeito em PE e em PB". Paper presented at the 3rd Workshop do Projecto PE-PB. Lisboa, September Erteschik-Shir, Nomi (1997). The Dynamics of focus structure. Cambridge:

33 Cambridge University Press Ferreira, Marcelo B. (2000) Argumentos Nulos em Português Brasileiro. Dissertação de Mestrado, UNICAMP, Campinas. Fiéis, Alexandra (2003). Ordem de Palavras, Transitividade e Inacusatividade. Reflexão Teórica e Análise do Português dos Séculos XIII a XVI, Dissertação de doutoramento, Universidade Nova de Lisboa Figueiredo Silva, Maria Cristina (1996). A posição do sujeito em português brasileiro - frases finitas e infinitivas, Campinas: Editora da Unicamp Galves, Charlotte (2001). Ensaios sobre as gramáticas do português. Campinas: Editora da Unicamp Kato, Mary & Fernando Tarallo (1986). Anything YOU can do in Brazilian Portuguese." Studies in Romance Linguistics, ed. Osvaldo Jaeggli and Carmen Silva-Corvalán. Dordrecht: Foris, Li, C-N.; S. Thompson. (1976) Subject and topic: a new typology of language. In: Li, C-N (ed.) Subject and Topic New York: Academic Press Newmeyer, Frederick (2007). In defense of the autonomy of syntax. Handout of presentation at Scandinavian Dialect Syntax Meeting, University of Tromsoe Pontes, E. (1987) O Tópico no Português do Brasil. Campinas: Ed. Pontes. Prince, Alan & Smolensky (1993). Optimality Theory: Constraint

34 Interaction in Generative Grammar. Rutgers University Center for Cognitive Science Technical Report 2 Ribeiro, Ilza (2008). O sujeito nulo referencial no português popular brasileiro séc. XIX, ms., Universidade Federal da Bahia Rizzi, Luigi (1986). On the Status of Subject Clitics in Romance. In Jaeggli e Silva-Corvalán, orgs, Studies in Romance Linguistics: Dordrecht: Foris. Silva, Cláudia Tavares (2004). A natureza de Agr e suas implicações na ordem VS: um estudo comparativo entre o português brasileiro e português europeu. Tese de doutoramento, Universidade Federal de Alagoas Vallduvi, Enric (1992) The Informational Component. New York: Garland